The Standish Group’s CHAOS report, published annually since 1996, gauges software development projects on the basis of on-time (schedule) and on-budget (cost) delivery that meets stakeholder needs. With success rates less than 50% for 25 years, pundits speculate on annual causes seeking to improve the percentages the following year.
Certainly the advent and institutionalizing of agile software development methods increased the figures from the dismal 35% statistics to the more palatable 43% project success, but through the years, little attention has been focused on one of the core reasons for not achieving on-time and on-budget delivery: the immaturity and unrealistic estimates themselves.
Given a propensity to be overly-optimistic, software teams over the years routinely under-estimate the effort, and duration of software development, while at the same time ignoring the risks and inherent uncertainties associated with the creative process. While it should be obvious that on-time and on-budget relies on realistic, historical data-based estimates, the situation is anything but.
Overly optimistic, unrealistic, and single value estimates (instead of a range) contribute to the poor-estimate problem, as does the lack of formal estimating processes. This is the topic of a series of upcoming blog posts based on observations gathered as the lead author for the International Cost Estimating and Analysis Association (ICEAA) Software Cost Estimating Body of Knowledge (CEBoK-S). When our industry increases the maturity of software cost estimating and takes a serious look at data-founded estimating, project success rates will increase. Join me on a software cost estimating journey.
Musings about Software Development 